REVIEWS The neuroscience of human intelligence differences Ian J. Deary, Lars Penke and Wendy Johnson Abstract | Neuroscience is contributing to an understanding of the biological bases of human intelligence differences. This work is principally being conducted along two empirical fronts: genetics — quantitative and molecular — and brain imaging. Quantitative genetic studies have established that there are additive genetic contributions to different aspects of cognitive ability — especially general intelligence — and how they change through the lifespan. Molecular genetic studies have yet to identify reliably reproducible contributions from individual genes. Structural and functional brain-imaging studies have identified differences in brain pathways, especially parieto-frontal pathways, that contribute to intelligence differences. There is also evidence that brain efficiency correlates positively with intelligence. Raven’s Progressive People differ along mental continua. Such individual and neuroscientific studies of general intelligence have Matrices test differences are the domain of differential psychology. yielded some clear results. Definitions of general intelli- An established non-verbal test Most research in this area of psychology focuses on cog- gence are shown in BOX 1. The terms (general) cognitive of inductive reasoning that is nitive and personality differences, which can be investi- ability, mental ability, intelligence and IQ (intelligence often regarded as a good marker of the general factor of gated as quantitative traits. Differential psychology has quotient) — in IQ’s lay and technical usages — are used intelligence. three main aims with respect to its traits of interest: to interchangeably to describe the strong common core that describe them accurately, to discover the real-life impact cognitive tests share. To illustrate the importance of scores Non-verbal reasoning of trait differences and to discover the aetiologies of trait on psychometric tests, we first describe their characteris- A broad subfactor of differences, including their biological bases. The field that tics and their impact on life. intelligence defined by tests that do not rely on verbal investigates the biological bases of individual differences Intelligence differences in the population approxi- stimuli or responses. The term in these traits is differential neuroscience. Here, we review mately follow a normal distribution, with the exception perceptual–organizational the differential neuroscience of human intelligence. of a slight excess at the lower end of the distribution ability is often used Individual differences in intelligence are usually caused by severe disorders that involve disrupted cogni- synonymously. measured using psychometric tests. These tests cover tive abilities. Males have a slight but consistently wider cognitive domains such as reasoning, processing speed, distribution than females at both ends of the range1. executive function, memory and spatial ability. Although Individual differences in human intelligence are among cognitive domains are sometimes considered to be the most robust observations in psychology. They are independent, differential psychology has firmly estab- quite stable in rank order throughout development2, lished that they are not: people who perform well in one and even over long time spans. A single 45-minute test domain also tend to perform well in the others. This is of general intelligence had a correlation of 0.63 (0.73 recognized in the term ‘general intelligence’, which is when disattenuated for restriction of range) in people 3 Centre for Cognitive Ageing usually designated ‘g’ (discussed below and in BOX 1). tested twice, at ages 11 and then 79 years . General intel- and Cognitive Epidemiology, Some individual tests — such as Raven’s Progressive ligence differences are associated with important life Department of Psychology, Matrices test, which is used to assess non-verbal reasoning outcomes, including school achievement4. In a study University of Edinburgh, Edinburgh EH4 2EE, — are good indicators of g. In this Review, we discuss involving tens of thousands of children, general intel- Scotland, UK. how neuroscience provides information about the ori- ligence at age 11 years had a correlation of over 0.8 with All authors contributed gins of differences in this general cognitive ability. scores on national tests of educational achievement equally to the work. We recognize that much of cognitive neuroscience 5 years later5. General intelligence is strongly predictive Correspondence to I.J.D. tends to focus on the cognitive domains themselves. of occupational attainment, social mobility6 and job e-mail: [email protected] 7 doi:10.1038/nrn2793 However, the neuroscientific aspect of general intelli- performance . People with higher general intelligence in Published online gence is important because general intelligence is respon- childhood or early adulthood have better health in middle 10 February 2010 sible for much of the predictive validity of cognitive tests, and later life, and are less likely to die young8. For example, NATURE REVIEWS | NEUROSCIENCE VOLUME 11 | MARCH 2010 | 201 © 2010 Macmillan Publishers Limited. All rights reserved REVIEWS Box 1 | Definitions of intelligence Each of these possibilities is correct to some degree, for the following reasons. First, scores on cognitive abil- An early and seemingly circular definition of intelligence came from the American ity tasks of all kinds are positively correlated. This is psychologist E. G. Boring in 1923, when he stated, “Intelligence is what the tests known as the positive manifold. In typical test batteries test”105. Although this definition is often criticized by detractors of IQ (intelligence consisting of 10–15 different cognitive tasks involving quotient)‑type tests, it was taken out of context. The apparently dismissive comment came after a summary of strong empirical findings — for example, that the tests a wide range of materials and content, a g factor almost showed marked individual differences, that the differences were stable over time, that always accounts for 40% or more of the total variance. children developed greater intelligence over time but tended to maintain the same Second, each individual cognitive test also shows a sub- rank order. The sentence immediately following the famous quote was that the famously stantial amount of more specific variance, generally glib definition “is only the point of departure for a rigorous discussion of the tests.” ranging from 20 to 50% of the total variance. Some of Boring was simply stating that the psychometric data had to be good and then linked this is attributable to error variance or variance resulting to other evidence about the origins and outcomes of intelligence. from factors such as fatigue and low mood and moti- A broader definition was agreed by 52 prominent researchers on intelligence: vation. However, some represents systematic variance “Intelligence is a very general capability that, among other things, involves the ability to specific to each test, and therefore reflects the particular reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly abilities involved in the test. Third, tests that are more and learn from experience. It is not merely book learning, a narrow academic skill, or test‑taking smarts. Rather, it reflects a broader and deeper capability for similar in content are more closely correlated with each comprehending our surroundings—‘catching on’, ‘making sense’ of things, or ‘figuring other than with tests that have different content. That is, out’ what to do. Intelligence, so defined, can be measured, and intelligence tests people tend to have areas of relative strength and weak- measure it well”106. ness in certain broad domains of cognitive ability. For Intelligence tests generally consist of either complex tasks that involve different example, some are very good at solving various problems aspects of reasoning, such as the Ravens Progressive Matrices, or batteries of tasks that involving spatial manipulation but not quite as good at require different kinds of cognitive performance, such as providing definitions of words verbal problems, whereas others show the opposite pat- or visualizing three‑dimensional objects from two‑dimensional diagrams. Two properties tern. These individual differences in broad cognitive of these kinds of tests are important. First, all intelligence tests — whether of single, domains — though given much attention in cognitive unitary tasks or complex, multi‑faceted tasks — are correlated and tend to generate a neuroscience — contribute a small amount of variance strong general factor when applied to a large sample of people. Second, whatever our definition, intelligence should be assessed by its construct validity, meaning the compared with g and the specific tests. Fourth, some of accumulated evidence that the tests measure something of relevance: evidence on the variance also reflects individual differences in expo- practical outcomes of intelligence differences, consistency of psychometric structure, sure to testing in general and exposure to the specific and relationships with biological structures and processes. By that criterion, intelligence tests involved in particular. is a core and valid facet of individual differences among humans. As this article shows, An example of how the hierarchical structure of intelli- irrespective of definition and test used, data from brain‑imaging